Expectancy and R-Multiples: Measuring Trading Edge Across a Portfolio

Direct Answer

An R-multiple measures a trade's result as a multiple of the amount it risked — a trade risking $200 that made $600 is +3R, one that hit its stop is exactly -1R. Expectancy in R equals (win rate × average win in R) minus (loss rate × average loss in R), and it is the average edge a system produces per trade over a large sample. Rolled up across a portfolio, blended expectancy is a risk-capital-weighted average of every strategy's own expectancy, not a simple average — and correlation between strategies shapes the ride to that average far more than the average itself does.

The practical objective is not to compute one number and stop. It is to understand what expectancy can and cannot tell a trader, calculate it correctly at both the single-strategy and portfolio level, and avoid treating a long-run average as a forecast for the next trade.

Key Takeaways

• An R-multiple normalizes a trade's result by the dollar amount it risked, not by the dollar amount it made or lost.
• Expectancy is a weighted average of win rate, average win size, and average loss size, all expressed in R.
• Portfolio-level expectancy blends each strategy's expectancy by risk-capital contribution, not by simple averaging.
• Correlation between strategies changes the variance and drawdown shape of the blended result, even when it does not change the blended mean.
• A positive expectancy is a long-run statement about many trades — it says nothing about any single trade's outcome.
• Expectancy computed from a small sample, or from R-multiples defined inconsistently across strategies, is unreliable.

What Is an R-Multiple?

An R-multiple restates a trade's profit or loss as a multiple of the dollar amount that trade risked, so results from different position sizes and different instruments can sit on the same scale.

The "R" stands for the initial risk on a trade — the dollar distance between the entry price and the stop-loss price, multiplied by position size. For a long position, the formula is:

R = (Exit Price − Entry Price) ÷ (Entry Price − Stop Price)

For a short position, the direction of both the numerator and denominator flips:

R = (Entry Price − Exit Price) ÷ (Stop Price − Entry Price)

Because the denominator is fixed the moment a trade is placed and a stop is set, R-multiples describe outcomes purely in terms of risk taken, independent of account size or position size. A trade risking $200 in planned loss that closes for a $600 profit is a +3R trade, regardless of whether that $200 came from 100 shares with a $2 stop or 10 shares with a $20 stop. A trade that is stopped out exactly at its planned stop price is, by construction, exactly -1R — the exit price equals the stop price, so the numerator equals the negative of the denominator. A trade that gets a worse fill than its stop during a gap or fast market can produce a result worse than -1R, which is one reason -1R is a planning assumption, not a guaranteed floor.

Practical checklist

• Define R at trade entry, using the initial stop-loss distance, before the trade is ever managed.
• Keep the R denominator fixed to the initial stop, even if the stop is later moved to reduce risk or lock in gains.
• Record R-multiples for every closed trade, not only the ones that turned out well.
• Use the same R formula, adjusted for direction, for both long and short trades.
• Treat a fill worse than the planned stop as a valid result below -1R, not as an error to discard.

Common mistake: recalculating a trade's R using a stop that was moved during the trade — such as trailing it to breakeven — rather than the original stop distance set at entry. This inflates the R-multiple of winners that were later protected and makes results across trades no longer comparable to each other.

The Expectancy Formula

Expectancy converts a distribution of individual R-multiples into a single average figure that describes the system's edge per trade.

Expectancy (in R) = (Win% × Average Win in R) − (Loss% × Average Loss in R)

Win% and Loss% are the fraction of trades that closed positive and negative, and they sum to 1 (ignoring the rare exact-breakeven trade). Average Win in R and Average Loss in R are the mean R-multiple of the winning trades and the mean R-multiple of the losing trades, with the loss expressed as a positive magnitude so the formula subtracts it. The result is the average R-multiple the system has produced per trade across the sample — not a prediction of the very next trade, but the rate at which edge, if any, accumulates over many trades.

A fully worked example

Consider a system with a 45% win rate, an average winning trade of +2.8R, and an average losing trade of -1R:

Expectancy = (0.45 × 2.8) − (0.55 × 1.0)
Expectancy = 1.26 − 0.55
Expectancy = +0.71R per trade

This means that, averaged across a large number of trades, this system should produce about 0.71R of profit per trade. It does not mean every trade nets 0.71R — most trades will be either a loss near -1R or a win somewhere around +2.8R, with 0.71R emerging only as the long-run average of that mix. To translate the figure into dollars, multiply it by the dollar amount risked per trade at a given position-sizing rule. Risking 1% of a $100,000 account means each trade risks $1,000, so 1R equals $1,000, and the expected value per trade is:

Expected value per trade = Expectancy (R) × Dollar risk per trade
Expected value per trade = 0.71 × $1,000 = $710

Over a large sample of trades at this position size, the system should average roughly $710 of profit per trade. That figure scales directly with the dollar risk per trade: doubling the risk-per-trade percentage to 2% doubles the expected dollar value per trade to roughly $1,420, but it also doubles the dollar size of the drawdowns the account will experience along the way — expectancy in R does not change with position sizing, but the dollar consequences of that expectancy do.

Practical checklist

• Calculate win rate, average win in R, and average loss in R from a consistent, complete trade sample.
• Express average loss in R as a positive number before subtracting it in the formula.
• Convert expectancy to dollars only by multiplying by the actual dollar risk per trade at the sizing rule in use.
• Recalculate expectancy periodically as more trades accumulate, rather than treating an early figure as final.
• Separate expectancy from trade frequency — a smaller per-trade expectancy taken often can outearn a larger one taken rarely.

Common mistake: reporting expectancy as a dollar figure without stating the position-sizing rule behind it, which makes the number impossible to compare against any other system or to reproduce at a different risk-per-trade setting.

Rolling Expectancy Up to the Portfolio Level

A portfolio running several strategies at once needs a blended expectancy figure, weighted by how much risk-capital each strategy actually contributes.

Because R already expresses a result as a fraction of the capital that specific trade risked, R-multiples from different strategies can be combined into a single blended expectancy by weighting each strategy's own expectancy by its share of total risk-capital deployed, rather than by simply averaging the strategies' expectancy figures together. A strategy that risks a larger share of the account's total risk budget pulls the blended figure toward its own expectancy more than a strategy contributing a smaller share, even if the smaller strategy has a higher standalone expectancy. This is why a portfolio's blended expectancy is very rarely identical to any single strategy's own number — it is a weighted mixture, and the weights matter as much as the individual figures.

Two complications make this rollup less mechanical than it first appears. The first is differing position-sizing rules: if one strategy risks 0.5% of equity per trade and another risks 2% per trade, then "1R" represents a very different dollar amount in each. Blending their raw expectancy figures without accounting for how much capital sits behind each R can misstate the portfolio's actual dollar-weighted edge — the correct approach weights by dollars of risk-capital contributed, not by trade count or by treating every strategy's R as equivalent. The second is correlation: the expected value of a sum of trade outcomes is the sum of their expected values regardless of how correlated those outcomes are with each other, so the blended mean expectancy itself is unaffected by correlation. What correlation changes is the shape of the path toward that mean. Two strategies that tend to win and lose together produce a portfolio equity curve with deeper, more concentrated drawdowns than the same two strategies would produce if their results were largely independent, even though the long-run blended expectancy figure is identical either way.

Practical checklist

• Weight each strategy's expectancy by its share of total risk-capital deployed, not by a simple average across strategies.
• Convert every strategy's R-multiples to a common risk-capital basis before blending, since different position-sizing rules make 1R mean different dollar amounts.
• Track correlation between strategies separately from the blended expectancy figure, since expectancy alone does not capture it.
• Expect the blended portfolio figure to sit between the best and worst individual strategy, not above the best one.
• Re-run the blended calculation whenever a strategy's risk-capital allocation changes materially.

Common mistake: averaging strategy expectancies with equal weight regardless of how much risk-capital each strategy actually uses, which overstates the influence of a small, high-expectancy strategy and understates the influence of the strategy actually carrying most of the portfolio's risk.

Why R-Multiples Beat Raw Dollars for Comparing Strategies

Dollar profit and loss figures conflate position size with the quality of the underlying edge, which R-multiples strip out.

A strategy that nets $50,000 in a year and a strategy that nets $5,000 in a year cannot be compared on those figures alone, since the first might be trading ten times the capital, or risking ten times as much per trade, to produce that larger number. Expressing results in R removes position size from the comparison entirely: a system generating +0.71R per trade has the same quality of edge whether it is run on a $10,000 account or a $10,000,000 account, because R is defined relative to the risk taken on each individual trade, not relative to account size. This is what makes expectancy in R useful for comparing a small-cap stock strategy using $0.10 stops against a futures strategy using $500 stops, or for comparing a strategy's live results against its backtested results run at a different size.

Dollar-denominated comparisons remain useful for a different question — how much a strategy actually contributed to total account growth — but that is a question about capital allocation, not about the quality of the edge itself. Treating a strategy's larger dollar profit as evidence of a "better" system, without normalizing for how much was risked to produce it, is a common way a genuinely weaker edge run at larger size gets mistaken for a stronger one.

Practical checklist

• Compare strategies using expectancy in R before comparing them using raw dollar profit and loss.
• Use dollar figures only to answer capital-allocation questions, not edge-quality questions.
• Recompute R-based comparisons if a strategy's typical stop distance or position size changes materially over time.
• Be explicit about account size and risk-per-trade whenever a dollar comparison is presented alongside an R comparison.
• Treat a larger dollar profit at a larger risk level as a capital-allocation outcome, not proof of a stronger edge.

Common mistake: concluding that a strategy with higher total dollar profit has "worked better" than one with lower total dollar profit, without checking whether the higher-profit strategy simply risked more capital per trade to get there.

Variance, Losing Streaks, and a Positive-Expectancy Edge

A genuine positive expectancy still permits long losing streaks, because expectancy describes an average, not a schedule.

Expectancy is the mean of a distribution of R-multiples, and any distribution with real variance can still produce extended runs below its own average purely by chance. A system with a 45% win rate and +0.71R expectancy will, over many samples, occasionally string together six, eight, or more consecutive losing trades even though nothing about the edge has broken down — that is simply what a distribution with a 55% chance of loss on any given trade will periodically produce. The statistical concept at work is variance around a positive mean: the long-run average is real, but the short-run path toward it can look nothing like the average, and there is no mechanism that forces the average to "catch up" on any particular timetable.

This is precisely why position sizing and bankroll discipline matter as much as expectancy itself. A system with strong positive expectancy can still be ruined by a losing streak if each trade risks too large a share of the account, because a string of -1R losses at 5% risk per trade erodes capital far faster, and requires a much larger percentage gain to recover from, than the same string of losses at 1% risk per trade. Sizing conservatively enough to survive the losing streaks a positive-expectancy system will periodically produce is what allows the long-run average to actually show up in the account's equity curve, rather than the account being forced out of the game by variance before the edge has had enough trades to express itself.

Practical checklist

• Expect losing streaks even from a system with confirmed positive expectancy, and size positions so a streak does not threaten the account.
• Distinguish a losing streak caused by normal variance from one caused by a genuine breakdown in the system's edge.
• Keep risk per trade low enough that several consecutive -1R losses remain a manageable, recoverable drawdown.
• Avoid increasing position size to "make back" losses from a recent losing streak, which compounds variance risk on top of itself.
• Review expectancy over a rolling window of trades to detect a real edge breakdown rather than reacting to any single streak.

Common mistake: abandoning a system, or overriding its signals, after a losing streak that is fully consistent with its known win rate and R distribution, mistaking ordinary variance for evidence the edge has stopped working.

Worked Portfolio Example

Assume a $200,000 account runs three strategies at once, each with its own measured expectancy and its own share of a monthly $50,000 risk-capital budget.

Inputs

• Strategy A, trend-following stocks: expectancy +0.71R, deploys $25,000 of risk-capital this month (50% of the total)
• Strategy B, crypto mean-reversion: expectancy +0.40R, deploys $15,000 of risk-capital this month (30% of the total)
• Strategy C, options income: expectancy +0.20R, deploys $10,000 of risk-capital this month (20% of the total)

Formula

Expected profit contribution = Risk-capital deployed × Strategy expectancy (R)
Strategy A: $25,000 × 0.71 = $17,750
Strategy B: $15,000 × 0.40 = $6,000
Strategy C: $10,000 × 0.20 = $2,000
Total expected profit = $25,750

Blended portfolio expectancy = Total expected profit ÷ Total risk-capital deployed
Blended portfolio expectancy = $25,750 ÷ $50,000 = +0.515R

The blended figure of +0.515R sits between the portfolio's strongest strategy (+0.71R) and its weakest (+0.20R), pulled toward Strategy A because it contributes the largest share of risk-capital. This is the expected value math, and it holds regardless of how correlated the three strategies are with each other — expectation adds linearly no matter what.

What correlation changes is not this $25,750 expected figure but the shape of the month that actually produces it. Strategy A (trend-following stocks) and Strategy C (options income) are both directional equity-market exposures and tend to move together — both are more likely to have a losing stretch during the same broad selloff. Strategy B (crypto mean-reversion) is largely uncorrelated with either. A month in which A and C both land losing stretches at the same time will show a rougher, more front-loaded drawdown than the blended math alone suggests, even though the expected $25,750 for the month hasn't changed; B's independence is what actually smooths that ride, not the expectancy calculation itself. Reviewing blended expectancy alongside a separate correlation check, rather than treating the expectancy figure as a complete risk picture on its own, is what catches this gap.

Misconceptions Versus Reality

Common Mistakes

Two mistakes account for most unreliable expectancy figures in practice, and both are easy to avoid once they're named.

Computing expectancy from too small a sample

Win rate and average R are themselves estimates, and estimates built from a small number of observations carry wide uncertainty. A system that shows +0.71R expectancy after 20 trades could plausibly be a system with a true expectancy anywhere from modestly negative to well above +1R — 20 trades simply is not enough to narrow that range meaningfully. Treating an early, small-sample figure as a settled fact leads to sizing decisions built on noise rather than edge. The fix is patience: track the figure on a rolling basis, watch how much it moves as new trades are added, and hold off on high-confidence conclusions until the sample runs well past 100 trades, ideally across more than one type of market condition.

Mixing R-multiples calculated with inconsistent stop-placement rules

Expectancy across a portfolio only means something if every strategy contributing to it defines R the same way. If one strategy calculates R from the initial stop set at entry and never revises it, while another recalculates R using a stop that was trailed during the trade, the two strategies' R-multiples are not measuring the same thing even though they look identical on a spreadsheet. Blending them produces a blended expectancy figure that appears precise but is actually comparing apples to a moving target. The fix is a single, written stop-placement convention applied identically across every strategy before any R-multiples are logged or combined — including how partial exits, scale-outs, and moved stops are handled — so the numbers being blended are actually the same kind of measurement.

A third, related mistake: selectively excluding trades from the sample — dropping outlier losses as "anomalies" or omitting trades that were manually overridden — which inflates the calculated expectancy above what the system actually produces when every trade is included honestly.

Risks, Limitations, and Exceptions

• Expectancy calculated from a small sample can look strongly positive or negative purely by chance, independent of the system's true long-run edge.
• A system's true expectancy can shift over time as market conditions change, so a historical figure is not a permanent guarantee.
• R-multiples assume the stop fills at its planned price; gaps and illiquid markets can produce results worse than -1R.
• Blended portfolio expectancy hides correlation effects that materially change drawdown depth and timing, even at an identical blended mean.
• Inconsistent stop-placement conventions across strategies or traders make R-multiples look comparable while actually measuring different things.
• Position sizing decisions built on expectancy still have to account for variance and the possibility of extended losing streaks.
• Expectancy in R does not capture trade frequency; a smaller per-trade expectancy taken often can outperform a larger one taken rarely.
• Selectively excluding trades from a sample, even unintentionally, inflates the calculated expectancy above the system's real performance.

Practical Implementation Checklist

1. Define R at entry for every strategy, using a single, written stop-placement convention applied consistently.
2. Log the R-multiple of every closed trade, including losses, overrides, and outlier results.
3. Calculate win rate, average win in R, and average loss in R once the sample is large enough to be meaningful.
4. Compute expectancy in R and convert it to dollars using the actual risk-per-trade percentage in use.
5. Determine each strategy's share of total risk-capital deployed before blending expectancy across the portfolio.
6. Calculate blended portfolio expectancy as a risk-capital-weighted average, not a simple average of strategies.
7. Check correlation between strategies separately, since it shapes drawdown risk that expectancy alone does not show.
8. Size positions conservatively enough to survive the losing streaks a positive-expectancy system will periodically produce.
9. Recalculate expectancy on a rolling basis as new trades accumulate, watching for genuine edge breakdown versus normal variance.
10. Record the review date, sample size, and inputs behind every expectancy figure so it can be checked later.

Tool Opportunity

A dedicated Swoopr tool should calculate per-trade R-multiples, single-strategy expectancy, and risk-capital-weighted portfolio expectancy automatically from logged trades.

Recommended inputs: entry price, stop price, exit price, and direction for every trade; the strategy each trade belongs to; the dollar risk-capital each strategy deploys over the review period; and account equity.

Expected outputs: per-trade R-multiple, rolling win rate and average win/loss in R, single-strategy expectancy in R and in dollars, blended portfolio expectancy weighted by risk-capital contribution, and a running sample-size indicator that flags when a figure is still statistically preliminary.

Validation requirements: reject trades missing a stop price, flag inconsistent stop-placement conventions across strategies, distinguish a fill worse than the planned stop from a standard -1R result, and never present a small-sample expectancy figure without a clear preliminary-data warning.

Frequently Asked Questions

Sources

• Van Tharp Institute — Tharp Think Trading Concepts, the primary source for the R-multiple and expectancy framework used throughout this page.
• CFA Institute — Portfolio Risk and Return: Part II, covering how correlation and weighting affect combined portfolio risk and return.
• CME Group Institute — The Benefits of Portfolio Diversification, on how correlation between positions or strategies affects diversification benefits.

Conclusion

An R-multiple measures a trade against what it risked, and expectancy averages those R-multiples into a single figure that says whether a system has a real edge over many trades.

Use this page as part of the larger Swoopr learning architecture. Move to the parent hub for the broader set of performance metrics, or to a supporting guide when a specific calculation, comparison, or workflow is required — the position-sizing rules behind R, the trading journal where R-multiples get logged, and win rate versus profit factor as a different lens on the same trade data.

Related Reading

• Portfolio performance metrics hub — the parent guide this page belongs to.
• Win rate vs. profit factor — a different framing of the same trade-outcome data, useful alongside expectancy rather than instead of it.
• Position sizing and risk per trade — where the initial dollar risk that defines 1R actually comes from.
• Trading journal — where R-multiples get logged trade by trade so expectancy can be calculated from real data.